HDR Merging of RAW Exposure Series for All-Sky Cameras: A Comparative Study for Circumsolar Radiometry

Area of Science:

• Solar energy meteorology
• Atmospheric optics
• Image processing

Background:

• All-sky imagers (ASIs) face extreme dynamic range challenges in solar energy meteorology.
• Current operational pipelines often use image signal processor (ISP)-based high-dynamic-range (HDR) methods, which can degrade image quality near the Sun.

Purpose of the Study:

• To contrast radiance-linear HDR fusion in the sensor/RAW domain (RAW-HDR) with vendor ISP-based HDR (ISP-HDR).
• To evaluate the impact of different HDR methods on circumsolar image quality and suitability for downstream solar energy applications.

Main Methods:

• Paired, interleaved image acquisitions under clear and broken-cloud conditions.
• Solar-based geometric calibration for Sun-centered analysis.
• Quantitative evaluation using saturated-area fraction and radial gradient metrics in the radiance-linear HDR domain.

Main Results:

• ISP-HDR showed approximately double the near-saturation within 0-4° of the Sun compared to RAW-HDR.
• ISP-HDR exhibited a three- to fourfold weaker circumsolar radial gradient within 0-6° compared to RAW-HDR.
• RAW-HDR better preserved circumsolar structure.

Conclusions:

• Radiance-linear fusion in the RAW domain (RAW-HDR) is superior to ISP-domain HDR for preserving circumsolar image detail.
• RAW-HDR provides more suitable input for critical downstream tasks like cloud-edge detection, aerosol retrieval, and irradiance estimation in solar energy meteorology.